Energy Optimization in B5G Networks with Predictive Allocation of Base Stations in Edge Server Placement

Modern network systems rely heavily on edge servers (ESs) to enhance performance and ensure reliability Beyond 5G (B5G) networks. ESs are crucial in minimising network latency, improving responsiveness, and optimising network utilisation by bringing computational resources closer to endusers. However, randomly connecting base stations (BSs) to ESs leads to inefficient ES utilisation and decreased efficiency. In contrast, strategically allocating BSs to ESs can significantly improve efficiency. This research investigates the importance of ESs in modern B5G network systems and their contribution to delivering efficient and reliable Quality of Service (QoS). A predictive allocation (PA) technique is employed to address this, leveraging predictive modelling to forecast future demand and allocate BSs to ESs accordingly. We consider the opensource Shanghai Telecom dataset to evaluate different allocation approaches and their impact on QoS. The evaluation focuses on essential QoS metrics such as workload balance, energy consumption, and server utilisation rate. We compare allocation approaches, including the Top-First, Random, and K-mean approaches. The results highlight the superiority of a strategic allocation approach empowered by PA. The proposed method outperforms the random allocation approach by achieving 13.08%, 22.92%, 18.77%, and 9.37% in terms of higher workload distribution, utilisation rate, energy consumption, and latency improvements.